Dynamic Interrelations Among Major World Stock Markets: A Neural Network Analysis
Bibliographic record
Abstract
ABSTRACT This paper investigates the application of artificial neural networks to the dynamic interrelations among major world stock markets. The database for this study consists of daily stock market indices of major world stock markets. These stock market indices are: Canada, France, Germany, Japan, United Kingdom (UK), the United States (US), and the world excluding US (World). Based on the criteria of Root Mean Square Error, Maximum Absolute Error, and the value of the objective function, it is found that Multilayer Perceptron models with logistic activation functions predict daily stock returns better than traditional Ordinary Least Squares and General Linear Regression models. Furthermore, it is found that a multilayer perceptron with five units in the hidden layer better predicts the stock indices for USA, France, Germany, UK and World than a neural network with two hidden elements. It is concluded that neural systems can be used as an alternative tool for financial analysis. JEL: C3, C32, C45, C5, C63, F3, G15 Keywords: Neural networks; Major stock markets; Dynamic interrelations; Forecasting I. INTRODUCTION Neural networks are powerful forecasting tools that draw on the most recent developments in artificial intelligence research. They are non-linear models that can be trained to map past and future values of time series data and thereby extract hidden structures and relationships that govern the data. Neural networks are applied in many fields such as computer science, engineering, medical and criminal diagnostics, biological investigation, and economic research They can be used for analysing relations among economic and financial phenomena, forecasting, data filtration, generating time-series, and optimization (Hawley, Johnson, and Raina, 1990; White, 1998; White 1996; Tema, 1997; Cogger, Koch and Lander, 1997; Cheh, Weinberg, and Yook, 1999; Cooper, 1999; Hu and Tsoukalas, 1999; Moshiri, Cameron, and Scuse, 1999; Shtub and Versano, 1999; Garcia and Gencay, 2000; and Hamm and Brorsen, 2000). This paper investigates the application of artificial neural networks to the dynamic interrelations among major world stock markets.' These stock market indices are: Canada, France, Germany, Japan, United Kingdom (UK), the United States (US), and the world excluding US (World). Based on the criteria of Root Mean Square Error (RMSE), Maximum Absolute Error (MAE), and the value of the objective function the model is compared to other statistical methods such as Ordinary Least Squares (OLS) and General Linear Regression Model (GLRM). Neural networks have found ardent supporters among various avant-garde portfolio managers, investment banks and trading firms. Most of the major investment banks, such as Goldman Sachs and Morgan Stanley, have dedicated departments to the implementation of neural networks. Fidelity Investments has set up a mutual fund whose portfolio allocation is based solely on recommendations produced by an artificial neural network. The fact that major companies in the financial industry are investing resources in neural networks indicates that artificial neural networks may serve as an important method of forecasting. Artificial neural networks are information processing systems whose structure and function are motivated by the cognitive processes and organizational structure of neuro-biological systems. The basic components of the networks are highly interconnected processing elements called neurons, which work independently in parallel (Consten and May, 1996). Synaptic connections are used to carry messages from one neuron to another. The strength of these connections varies. These neurons store information and learn meaningful patterns by strengthening their inter-connections. When a neuron receives a certain number of stimuli, and when the sum of the received stimuli exceeds a certain threshold value, it fires and transmits the stimulus to adjacent neurons (Soh1, 1995). …
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".